Herbie

I open my eyes Saturday morning to see my son, Dave, dressed in a Boy Scout uniform. He is shaking my arm. Davey what are you doing here? I ask. He says, Dad, its seven oclock! Seven oclock? Im trying to sleep. Well be late, he says. WE will be late? For what? For the overnight hike! he says. Remember? You promised me I could volunteer you to go along and help the troopmaster. An hour and ten minues later, my son and I arrive at the edge of some forest. Waiting for us is the troop: 15 boys outfitted in caps, neckerchiefs, merit badges, the works. Well, at first Im a little mad at having all this foisted upon me. But then the idea of having to supervise a bunch of kids doesnt daunt me after all, I do that every day at the plant. So I gather everyone around. We look at a map and discuss objectives for this expedition into the perilous wilderness before us. The plan I learn, is for the troop to hike through the forest following a blazed trail to someplace called Devils Gulch, 10 miles away There we are to bivouac for the evening. In the morning we are to break camp and make our way back to the point of departure. So I line up the troop. Map in hand, I put myself at the front of the line in order to lead the way, and off we go. The weather is fantastic. The sun is shining through the trees. The skies are blue. Its breezy and the temperature is a little on the cool side, but once we get into the woods, its just right for walking. The trail is easy to follow because there are blazes (splotches of yellow paint) on the tree trunks every 10 yards or so. I suppose Im walking at about 2 miles/hour, which is about how fast the average person walks. At this rate, I think to myself, we should cover 10 miles in about 5 hours. My watch tells me it is 8:30 am. Allowing an hour and a half for breaks and lunch, we should arrive at Devils Gulch by 3 oclock, no sweat. After a few minutes, I turn and look back. The column of scouts has spread out to some degree frm the close spacing we started with. Instead of a yard or so between the boys, there are now larger gaps, some a little larger than others. I keep walking. I look back again after a few hundred yards, and the column is stretched out much farther, and a couple of big gaps have appeared. I can barely see the kid at the end of the line. I decide its better if Im at the end of the line instead of at the front. That way I know Ill be able to keep an eye on the whole column, and make sure nobody gets left behind. So I wait for the first boy to catch up to me, and I ask him his name. Im Ron, he says. Ron, I want you to lead the column, I tell him, handing over the map. Just keep following this trail, and set a moderate pace. Okay? Right, Mr. Rogo.

Everybody stay behind Ron! I call back to the others. Nobody passes Ron, because hes got the map. Understand? Everybody nods, waves. Everybody understands. I wait by the side of the trail as the troop passes. My son, Dave, goes by talking with a friend who walks close behind him. Five or six more come along, all of them keeping up without any problems. Then there is a gap, followed by a couple more scouts. After them, another, even larger gap has occurred. I look down the trail and I see this fat kid. He already looks a little winded. Behind him is the rest of the troop. Whats your name? I ask as the fat kid draws closer. Herbie, says the fat kid. You okay, Herbie? Oh, sure, Mr. Rogo, says Herbie. Boy its hot out, isnt it? Herbie continues up the trail and the others follow. Some of them look as if theyd like to go fastre, but they cant get around Herbie. I fall in behind the last boy. The line stretches out in front of me, and most of the time, unless were going over a hill or around a sharp bend in the trail, I can see everybody. The column seems to settle into a comfortable rhythm. .I begin to think about the conversation I had with Jonah in New York. I havent had any time to think about that. I did not understand what he was trying to make out of those two items he described. I mean, dependent events.statistical fluctuations so what? Theyre both quite mundane. Obviously we have dependent events in manufacturing. All it means is that one operation has to be done before a second operation can be performed. Parts are made in a sequence of steps. Machine A has to finish step 1 before worker B can proceed with step 2. All the parts have to be finished before we can assemble the product. The product has to be assembled before we can ship it and so on. But you find dependent events in any process, and not just those in a factory. Driving a car requires a sequence of dependent events. So does the hike were taking now. In order to arrive at Devils Gulch, a trail has to be walked. Up front, Ron has to walk the trail before Davey can walk it. Davey has to walk the trail before Herbie can walk it. In order for me to walk the trail, the boy in front of me has to walk it first. Its a simple case of dependent events. And statistical fluctuations? I look up and notice that the boy in front of me has going a little faster than I have been. Hes a few feet farther ahead of me than he was a minute ago. So I take some bigger steps to catch up. Then, for a second, Im too close to him, so I slow down. There: if Id been measuring my stride, I would have recorded statistical fluctuations. Whats the big deal? If I say that Im walking at the rate of 2 miles/hour I dont mean Im walking exactly at a constant rate of two miles per hour every instant. Sometimes Ill be going 2.5 miles/hour; sometimes maybe Ill be walking at only 1.2 miles/hour. The rate is going to fluctuate according to the length and speed of each step, but over time and distance, I should be averaging about 2 miles/hour, more or less. The same thing happens in the plant. How long does it take to solder the wire leads on a transformer? Well if you get out your stopwatch and time

the operation over and over again, you might find that it takes, lets say, 4.3 minutes on the average, but the actual time on any given instance may range between 2.1 minutes up to 6.4 minutes. Nobody in advance can say, This one will take 2.1 minutesthis one will take 5.8 minutes. Nobody can predict that information. So whats wrong with that? Nothing as far as I can see. Anyway, we dont have any choice. What else are we going to use in place of an average or an estimate. I find Im almost stepping on the boy in front of me. Weve slowed down somewhat. Its because were climbing a long, fairly steep hill. All of us are backed up behind Herbie. Atta boy, Herbie! I say to encourage him. Lets keep it moving! Herbie his face red from the climb disappears over the crest. The others continue the climb, and I trudge behind them until I get to the top. Pausing there, I look down the trail. Holy cow! Wheres Ron? He must be half a mile ahead of us. I can see a couple of boys in front of Herbie, and everyone else is lost in the distance. I cup my hands over my mouth and yell HEY! LETS GO UP THERE! LETS CLOSE RANKS! DOUBLE TIME! DOUBLE TIME! Herbie eases into a trot. The kids behind him start to run. I jog after them. After a couple of hundred yards we still havent caught up. Herbie is slowing down. The kids are yelling at him to hurry up. Im huffing and puffing along. Finally I can see Ron off in the distance. HEY RON! I shout. HOLD UP! The call is relayed up the trail by the other boys. Ron, who probably heard the call the first time, turns and looks back. Herbie, seeing relief in sight, slows to a fast walk. And so do the rest of us. Ron, I thought I told you to set a moderate pace, I say. But I did! he protests. Well, lets just all stay together next time, I tell them. We start out again. The trail is straight here, so I can see everyone. We havent gone thirty yards before I notice it starting all over again. The line is spreading out; gaps between the boys are widening. Dammit, were going to be running and stopping all day long if this keeps up. Half the troop is liable to get lost if we cant stay together. Ive got to put an end to this. The first one I check is Ron. Ron though, is indeed setting a steady, average pace for the troop a pace nobody should have any trouble with. I look back down the line, and all of the boys are walking at about the same rate as Ron. Herbie? Hes not the problem anymore. Maybe he felt responsible for the last delay, because now he seems to be making a special effort to keep up. If we are all walking at about the same pace, why is the distance between Ron, at the front of the line, and me, at the end of the line, increasing? Statistical fluctuations? Nah, couldnt be. The fluctuations should be averaging out. We are all moving at about the same speed, so that should mean the distance between any of us will vary somewhat, but will even out over a period of time. The distance between Ron and me should also expand

and contract within a certain range, but should average out about the same throughout the hike. But it isnt. As long as each of us is maintaining a normal moderate pace like Ron, the length of the column is increasing. The gaps between us are expanding. Why arent all of us being able to walk the same pace as Ron and stay together? Im watching the line when something up ahead catches my eye. I see Davey slow down for a few seconds. Hes adjusting his packstraps. In front of him, Ron continues onward, oblivious. A gap of tenfifteentwenty feet opens up. The entire line has grown by 20 feet. Thats when I begin to understand whats happening. Ron is setting the pace. Every time someone moves slower than Ron, the line lengthens. It wouldnt even have to be as obvious as when Dave slowed down. If one of the boys takes a step thats half an inch shorter than the one Ron took, the length of the whole line could be affected. But what happens when someone moves faster than Ron? Arent the longer or faster steps supposed to make up for the slower ones? Dont the differences average out? Suppose I walk faster. Can I shorten the length of the line? Between me and the kid ahead of me is a gap of about 5 feet. If he continues walking at the same rate, and if I speed up, I can reduce the gap and maybe reduce the total length of the line. However, I can only do that until Im bumping the kids rucksack (and if I did that hed sure as hell tell his mother). So I have to slow down to his rate. Ince Ive closed the gap between us, I cant go any faster than the rate at which the kid in front of me is going, and he in turn cant go any faster than the kid in front of him and so on up the line to Ron. Our speeds depend upon the speeds of those in front of us in the line. Its starting to make sense. Our hike is a set of dependent eventsin combination with statistical fluctuations. Each of us is fluctuating in speed, faster and slower. Even if I could walk 5 miles/hour, I couldnt do it if the boy in front of me could only walk 2 miles/hour. Even if the kid directly in front of me could walk that fast, neither of us could do it unless all the boys in the line were moving at 5 miles/hour at the same time. So Ive got limits on how fast I can go. However, there is no limit on my ability to slow down. Or on anyones ability to slow down. Or stop. If any of us did the line would extend indefinitely. Whats happening isnt an averaging out of the fluctuations in our various speeds, but an accumulation of the fluctuations. Mostly it is an accumulation of slowness because dependency limits the opportunities for higher fluctuations. That is why the line is spreading. We can make the line shrink only by having everyone in the back of the line move much faster than Rons average over some distance. To make the total length of the line contract, I have to move faster than average for a distance equal to all the excess space between all the boys. I have to make up for the accumulation of all their slowness. This troop of boys is analogous to a manufacturing systemsort of a model. There are both dependent events and statistical fluctuations. In fact, the troop does produce a product; we produce walked trail. Ron begins production by consuming the unwalked trail before him, which is the

equivalent of raw materials. Ron processes the trail first by walking over it, then Davey has to process it next, followed by the boy behind him, and so on back to Herbie and the others and on to me. Each of us is like an operation which has to be performed to produce a product in the plant; each of us is one of a set of dependent events. Does it matter what order we are in? Well somebody has to be first and somebody else has to be last, so we have dependent events no matter if we switch the order of the boys. I am the last operation. Ony after I have walked the trail is the product sold so to speak. That would have to be our throughput, not the rate at which Ron walks the trail, but the rate at which I do. What about the amount of trail between Ron and me? It has to be inventory. Ron is consuming raw materials, so the trail the rest of us are walking is inventory until it passes behind me. And what is operational expense? Its whatever lets us turn inventory into throughput, which in our case qould be the energy the boys need to walk. I cant really quantify that for the model, except that I know when Im getting tired. If the distance between Ron and me is expanding, it can only mean that inventory is increasing. Throughput is my rate of walking. Which is influenced by the fluctuating rates of the others. So as the slower than average fluctuations acumulate, they work their way back to me. Which means that, relative to the growth of inventory, throughput for the entire system goes down. Statistical Fluctuation and Dependence Compare the hike to most production and service processes. If there is variation within and between the steps in the processes in your organization, then you have statistical fluctuation. This problem is well discussed in the quality literature and reduction in this fluctuation is a necessary condition for just-in-time inventory systems. If the steps in the process must be done in a particular order then the process steps have dependence. Wood must be sanded before it is painted, the value of an insurance claim must be determined before the claim check is processed, and patient insurance status should be confirmed before a hospital room is assigned. The flow of work into any station depends on the timely completion of the work in previous stations. This is the narrow trail / no passing assumption in our analogy. In order for any scout to move forward, the scout in front of him has to have already moved forward. Steps that are later in the process must wait for all earlier steps to be completed. Five Steps Here are the steps in the improvement process 1. Identify the system's constraints. 2. Decide how to exploit the system's constraints. 3. Subordinate everything else to the above decision. 4. Elevate the system's constraints 5. If a constraint has been broken, go back to step 1.

Some Possible Solutions Herbie Leads If you want the scouts to arrive together and to arrive as quickly as possible, one possible alternative is to have Herbie lead the troop. When Herbie is slower than his average he slows down the arrival by exactly as much as in the previous example. When he moves faster than his average speed, the whole troop can move faster because all of the scouts behind him can catch up, at least eventually. With no one in front of Herbie to constrain his performance, he is able to move at his average speed. Production Schedule (Drums) There are steps in most processes that depend on other processes having already been completed. If there is dependence then Herbie cannot be moved to the front of the line. The operations that follow the constraint should have enough slack capacity to keep up with the pace of the constrained resource. The problem lies with the operations that precede the constraint. The scouts that are ahead of Herbie are not constrained by his slow pace. They are free to proceed at their own pace. Like the fast scouts at the front of the line who move ahead and produce gaps, these operations will move ahead and produce excess WIP inventory. The carrying cost on this inventory increases OE, reduces NI, ROI and CF. A possibility for keeping the front of the line from running away from the rest of the operations is, in effect, a drum beating a cadence that the constraint can keep up with. The drumbeat in a manufacturing setting is the production schedule, which dictates when and what material is supposed to be processed by what resource. Once you realize that the troop cannot move faster than the constraint, it becomes obvious that the production schedule must be dictated by the abilities of the constrained resource. If every production resource can produce to match the production schedule, then this system should work. If they cannot, then there will either be delays that reduce throughput or increases in work in process inventory. Sometimes the drum system is used in a push inventory system, where raw materials are released into the system to keep workers busy and keep efficiencies high on each resource in the system. In this case the cadence is not set to the slowest worker, but is set at or above the average. Expeditors and additional managerial attention are often needed to push work through the slower workstations. You can think of this as a Just-in-Case system. Work in Process inventory is high so that down time on any portion of the system does not endanger current production. High levels of inventory reduce the company's ability to respond to changing customer demands. While reported efficiencies are high, this system has a detrimental effect on NI, ROI, CF and OE, and therefore threatens future Throughput. Assembly Lines, Balanced Lines and JIT (Ropes)

The next possibility is a rope connecting each scout. This is effectively what you get with an assembly line or a production line where attempts have been made to balance the capacity of each workstation. The speed of the line beats the cadence and the structure of the line connects the workers to each other. The same system also describes just in time inventory systems. Here the cadence is set by market demand for finished goods. Transfer batch sizes are low, as is WIP inventory. Since production is driven by demand rather than warehouse capacity, the material produced is Throughput and not stored finished goods. The problem with this type of system is the existence of statistical fluctuations. With minimal work in process, any problem that occurs at any point in the process can bring the entire system to a halt. Successful JIT systems often require years of work to reduce the variability that naturally occurs in the process. While this may have a focusing effect for managers interested in solving production problems, it has a potentially devastating effect on current throughput. Drum-Buffer-Rope (DBR) What we need is a system that has low inventory and avoids downtime. The Theory of Constraints literature suggests a system called Drum-Buffer-Rope (DBR). In this system Herbie sets the cadence in that the production schedule is determined with the goal of matching the capacity of the constrained resource. The rope connects the constrained resource to the first resource in the process. The production schedule releases material to the first operation at exactly the rate that the constrained resource can process it. Therefore neither the first resource nor any other resource that precedes the constraint can produce excess inventory. A problem with any of these workstations in a Just-in-Time system causes Herbie to shut down and reduces throughput for the system. What is needed is a buffer of work in process inventory that will allow the earlier processes to catch up before the constraint runs out of work. Lengthening the rope connecting the constraint to the first process creates the buffer. These processes are constrained to run at the same speed as Herbie but are allowed to run slightly ahead. Downtime on processes that follow the constraint is not a problem because they have sufficient excess capacity to catch up with the constrained resource. There should be no build up of excess WIP inventory for these processes, again because they have more than enough capacity to process everything that Herbie hands them. A Simple Example Consider a process that has three sequential steps performed by three departments. Step A requires 9 minutes to complete. Step B requires 10 minutes. Step C requires 8 minutes. The capacities of the departments are

6.67, 6, and 7.5 units per hour respectively. What is the capacity of the organization? It is the capacity of the slowest department. Improving the productivity of either department A or C has no effect on organizational capacity. Within limits, down time and idle time in departments A and C have no impact on capacity. On the other hand, every minute of downtime or idle time in department B reduces throughput by 1/10th of a unit. Idle time in B reduces productive capacity for the entire system. First suppose that the system in place is the simple drum and the cadence is set at the average capacity of the three machines, around 6.7. Department A will be working at capacity. Inventory will pile up in department B's receiving area at a rate of 0.67 units per hour. Department C will process everything that they receive, but will be in trouble with management for low efficiencies (6/7.5 = 80% of capacity). Suppose we correct the cadence and set it at the capacity of B. This process is now running as a balanced assembly line with inventory arriving just in time. Every 10 minutes A receives enough raw materials to make one unit. A inspects the unit in the last few seconds of the process, and hands the unit to B who processes it and hands it to C. What happens if A discovers that they have produced a defective unit? B and C have to wait while A produces a replacement. We have lost 9 minutes of capacity for the entire plant. We will have a similar problem if a supplier delivers the material late, if A's equipment has a mechanical failure, if A is late getting back from lunch, or if natural variation causes A to produce at a rate slower than average for any reason. Based on historical information, we can estimate the magnitude of problems that department A might encounter, and we can calculate how much work in process inventory department B needs to have in their receiving area to sustain them while A catches up. For this example assume that 3 units are sufficient for A to catch up from any problems that they might encounter. We would allow department A to produce at their maximum capacity until they had accumulated 3 units of work in process. At that point they would have filled the buffer and hit the end of their rope. They would now receive only enough raw materials to produce one unit every 10 minutes. Whenever the buffer falls below the desired level, material is released to allow department A to replenish it to a safe level of work in process. Conclusion Both a pure drum system and a pure rope system have the same goal: the efficient operation of the system. They differ in their underlying assumptions. The pure drum system wants WIP inventory to protect against down time. Inventory is pushed into the system to try to keep all resources engaged. Each step in the process tries to operate at its locally optimal level. The pure rope system pulls inventory through, and seeks to reduce WIP inventory to eliminate potential quality and dependability problems hidden in large inventory pools, and to eliminate costs associated with carrying inventory and with slow response times to changing customer demands. The Theory of

Constraints approach accomplishes the goals of each of these systems using the Drum-Buffer-Rope method. Inventory is accumulated where it is needed to avoid system down time, and eliminated where it is not needed to lower cost and improve response to customer demand